Influence of Transfer Entropy in the Short-Term Prediction of Financial Time Series Using an ∊-Machine.

Area of Science:

• Quantitative Finance
• Information Theory
• Econometrics

Background:

• Financial time series prediction relies heavily on historical price data, influenced by various internal and external factors.
• Existing prediction models, such as epsilon-machines, utilize this history but can face challenges with prediction ties.
• Understanding inter-series dependencies is crucial for enhancing predictive accuracy in financial markets.

Purpose of the Study:

• To propose and evaluate a novel method for financial time series prediction incorporating entropy transfer.
• To investigate the influence of one financial series on another through information transfer.
• To refine epsilon-machine predictions by using entropy transfer to resolve ties.

Main Methods:

• Utilized epsilon-machine framework for financial time series prediction.
• Introduced and applied the concept of entropy transfer to quantify inter-series information flow.
• Analyzed six major financial indices: S&P 500, Nasdaq, Hang Seng, Nikkei 225, CAC 40, and DAX.

Main Results:

• Demonstrated that the prediction of a financial series' closing value can be influenced by known values of another series.
• Quantified information transfer across different global markets.
• Identified the S&P 500 and Nasdaq as the highest information-transferring series, followed by DAX/CAC 40, and then Nikkei 225/Hang Seng.

Conclusions:

• Entropy transfer is a viable method to enhance financial time series prediction accuracy.
• Inter-market dependencies, particularly from US markets, significantly impact other global financial series.
• The proposed method offers a robust approach to modeling complex financial market interactions.